Evaluation and selection of CORDEX-SA datasets and bias correction methods for a hydrological impact study in a humid tropical river basin, Kerala

It is well recognised that the performance of climate model simulations and bias correction methods is region specific, and, therefore, careful validation should always be performed. This study evaluates the performance of five general circulation model-regional climate model (GCM-RCM) combinations selected from CORDEX-SA datasets over a humid tropical river basin in Kerala, India, for climate variables such as precipitation, maximum and minimum temperatures. This involves ranking of the selected climate models based on an EDAS (Evaluation Based on Distance from Average Solution) method and the selection of an appropriate bias correction method for the selected three climate variables. A range of indices are used to evaluate the performance of the bias-corrected climate models to simulate observed climate data. Finally, the hydrological impact of the bias-corrected ranked models is assessed by simulating streamflow over the river basin using individual models and different combinations of models based on rank. According to the findings, hydrological simulation using an average of all GCM-RCM pairs provides the best model output in simulating streamflow, with an NSE value of 0.72. The results confirm the importance of a multimodel ensemble for improving the reliability and minimising the uncertainty of climate predictions for impact studies.

Automated summary

This study evaluated the performance of five GCM-RCM combinations over a humid tropical river basin in Kerala, India, ranking the climate models based on an EDAS method and selecting appropriate bias correction methods. The findings showed that using the average of all GCM-RCM pairs provided the best model output for simulating streamflow.